Merge branch 'united' into gui-and-scripting

aiserver.py

@@ -82,6 +82,7 @@ class vars:
     submission  = ""     # Same as above, but after applying input formatting
     lastctx     = ""     # The last context submitted to the generator
     model       = ""     # Model ID string chosen at startup
+    model_type  = ""     # Model Type (Automatically taken from the model config)
     noai        = False  # Runs the script without starting up the transformers pipeline
     aibusy      = False  # Stops submissions while the AI is working
     max_length  = 1024    # Maximum number of tokens to submit per action
@@ -391,9 +392,29 @@ if(not vars.model in ["InferKit", "Colab", "OAI", "ReadOnly", "TPUMeshTransforme
     vars.allowsp = True
     # Test for GPU support
     import torch
+    
+    # Make model path the same as the model name to make this consistent with the other loading method if it isn't a known model type
+    # This code is not just a workaround for below, it is also used to make the behavior consistent with other loading methods - Henk717
+    if(not vars.model in ["NeoCustom", "GPT2Custom"]):
+        vars.custmodpth = vars.model
+    # Get the model_type from the config or assume a model type if it isn't present
+    from transformers import AutoConfig
+    try:
+        model_config = AutoConfig.from_pretrained(vars.custmodpth)  
+    except ValueError as e:
+        vars.model_type = "not_found" 
+    if(not vars.model_type == "not_found"):
+        vars.model_type = model_config.model_type
+    elif(vars.model == "NeoCustom"):
+        vars.model_type = "gpt_neo"
+    elif(vars.model == "GPT2Custom"):
+        vars.model_type = "gpt2"
+    else:
+        print("WARNING: No model type detected, assuming Neo (If this is a GPT2 model use the other menu option or --model GPT2Custom)")
+        vars.model_type = "gpt_neo"
     print("{0}Looking for GPU support...{1}".format(colors.PURPLE, colors.END), end="")
     vars.hascuda = torch.cuda.is_available()
-    vars.bmsupported = vars.model in ("EleutherAI/gpt-neo-1.3B", "EleutherAI/gpt-neo-2.7B", "EleutherAI/gpt-j-6B", "NeoCustom")
+    vars.bmsupported = vars.model_type in ("gpt_neo", "gptj")
     if(args.breakmodel is not None and args.breakmodel):
         print("WARNING: --breakmodel is no longer supported. Breakmodel mode is now automatically enabled when --layers is used (see --help for details).", file=sys.stderr)
     if(args.breakmodel_layers is not None):
@@ -871,12 +892,13 @@ if(not vars.model in ["InferKit", "Colab", "OAI", "ReadOnly", "TPUMeshTransforme
             if("/" not in vars.model and vars.model.lower().startswith("gpt2")):
                 lowmem = {}
             
-            # Is CUDA available? If so, use GPU, otherwise fall back to CPU
+            # Download model from Huggingface if it does not exist, otherwise load locally
             
             if(os.path.isdir(vars.model.replace('/', '_'))):
                with(maybe_use_float16()):
                    tokenizer = GPT2TokenizerFast.from_pretrained(vars.model.replace('/', '_'), cache_dir="cache/")
                    model = AutoModelForCausalLM.from_pretrained(vars.model.replace('/', '_'), cache_dir="cache/", **lowmem)
+                   
             else:
                 print("Model does not exist locally, attempting to download from Huggingface...")
                 tokenizer = GPT2TokenizerFast.from_pretrained(vars.model, cache_dir="cache/")
@@ -1961,8 +1983,8 @@ def loadsettings():
 #  Allow the models to override some settings
 #==================================================================#
 def loadmodelsettings():
-    if(path.exists(vars.custmodpth + "/config.json")):
+    if(path.exists(vars.custmodpth.replace('/', '_') + "/config.json")):
-        model_config = open(vars.custmodpth + "/config.json", "r")
+        model_config = open(vars.custmodpth.replace('/', '_') + "/config.json", "r")
         js   = json.load(model_config)
         if("badwordsids" in js):
             vars.badwordsids = js["badwordsids"]

commandline.bat

@@ -1,15 +1,19 @@
 @echo off
-cd %~dp0
+cd /D %~dp0
 TITLE CMD for KoboldAI Runtime
 SET /P M=<loader.settings
 IF %M%==1 GOTO drivemap
 IF %M%==2 GOTO subfolder
 
 :subfolder
+SET TEMP=%~DP0MINICONDA3
+SET TMP=%~DP0MINICONDA3
 call miniconda3\condabin\activate
 cmd /k
 
 :drivemap
 subst K: miniconda3 >nul
+SET TEMP=K:\
+SET TMP=K:\
 call K:\python\condabin\activate
 cmd /k

install_requirements.bat

@@ -12,8 +12,7 @@ echo.
 SET /P B=Type the number of the desired option and then press ENTER: 
 
 Reg add "HKLM\SYSTEM\CurrentControlSet\Control\FileSystem" /v "LongPathsEnabled" /t REG_DWORD /d "1" /f 2>nul
-%~d0
+cd /D %~dp0
-cd %~dp0
 
 if exist miniconda3\ (
   echo Delete existing installation?
@@ -43,21 +42,27 @@ echo 1 > loader.settings
 subst K: /D >nul
 mkdir miniconda3 
 subst K: miniconda3
+SET TEMP=K:\
+SET TMP=K:\
 copy umamba.exe K:\umamba.exe
 K:
 umamba.exe create -r K:\python\ -n base
-IF %B%==1 umamba.exe install --no-shortcuts -r K:\python\ -n base -f "%~dp0\environments\huggingface.yml" -y
+IF %B%==1 umamba.exe install --no-shortcuts -r K:\python\ -n base -f "%~dp0\environments\huggingface.yml" -y --always-copy
-IF %B%==2 umamba.exe install --no-shortcuts -r K:\python\ -n base -f "%~dp0\environments\finetuneanon.yml" -y
+IF %B%==2 umamba.exe install --no-shortcuts -r K:\python\ -n base -f "%~dp0\environments\finetuneanon.yml" -y --always-copy
 umamba.exe -r K:\ clean -a -y
+rd K:\Python\pkgs /S /Q
 subst K: /d
 pause
 exit
 
 :subfolder
 echo 2 > loader.settings
+SET TEMP=%~DP0MINICONDA3
+SET TMP=%~DP0MINICONDA3
 umamba.exe create -r miniconda3\ -n base
-IF %B%==1 umamba.exe install --no-shortcuts -r miniconda3 -n base -f environments\huggingface.yml -y
+IF %B%==1 umamba.exe install --no-shortcuts -r miniconda3 -n base -f environments\huggingface.yml -y --always-copy
-IF %B%==2 umamba.exe install --no-shortcuts -r miniconda3 -n base -f environments\finetuneanon.yml -y
+IF %B%==2 umamba.exe install --no-shortcuts -r miniconda3 -n base -f environments\finetuneanon.yml -y --always-copy
 umamba.exe clean -a -y
+rd miniconda3\Python\pkgs /S /Q
 pause
 exit

play.bat

@@ -1,18 +1,23 @@
 @echo off
-%~d0
+cd /D %~dp0
-cd %~dp0
 TITLE KoboldAI - Server
 SET /P M=<loader.settings
 IF %M%==1 GOTO drivemap
 IF %M%==2 GOTO subfolder
 
 :subfolder
+ECHO Runtime launching in subfolder mode
+SET TEMP=%~DP0MINICONDA3
+SET TMP=%~DP0MINICONDA3
 call miniconda3\condabin\activate
 python aiserver.py %*
 cmd /k
 
 :drivemap
+ECHO Runtime launching in K: drive mode
 subst K: miniconda3 >nul
+SET TEMP=K:\
+SET TMP=K:\
 call K:\python\condabin\activate
 python aiserver.py %*
 subst K: /D

update-kobold.bat

@@ -7,11 +7,15 @@ IF %M%==1 GOTO drivemap
 IF %M%==2 GOTO subfolder
 
 :subfolder
+SET TEMP=%~DP0MINICONDA3
+SET TMP=%~DP0MINICONDA3
 call miniconda3\condabin\activate
 GOTO GIT
 
 :drivemap
 subst K: miniconda3 >nul
+SET TEMP=K:\
+SET TMP=K:\
 call K:\python\condabin\activate
 GOTO GIT